Image positioning device and method

Abstract

The present invention provides an image positioning device and method. The image positioning device comprises a calculation circuit and a positioning circuit. The calculation circuit obtains a plurality of images from an image extraction device, and classifies the plurality of images into a plurality of main groups according to a first algorithm, wherein each main group corresponds to a different one of a plurality of regions. Furthermore, the calculation circuit classifies the images included in each main group into a plurality of subgroups according to a feature of each image included in the main group and a subgrouping algorithm. The positioning circuit positions each subgroup in the region corresponding to the main group according to a relative positional relationship of each subgroup included in the main group.

Description

Technical Field

[0001] The embodiments of the present invention mainly relate to an image positioning technology, and more particularly to an image positioning technology for performing image positioning based on grouped images. Background Art

[0002] With the advancement of technology, the application of positioning is becoming more and more extensive. Therefore, how to produce more accurate image positioning results will be a topic worth studying. Summary of the Invention

[0003] In view of the above problems in the prior art, embodiments of the present invention provide an image positioning device and method.

[0004] According to an embodiment of the present invention, an image positioning device is provided.

[0005] The image positioning device includes a computing circuit and a positioning circuit. The computing circuit acquires multiple images from an image capture device and, based on a first algorithm, classifies the images into multiple main groups, each of which corresponds to a different one of multiple regions. Furthermore, the computing circuit classifies the images within each main group into multiple subgroups based on the characteristics of each image within the main group and a clustering algorithm. The positioning circuit is coupled to the computing circuit. Based on the relative positional relationships of the subgroups within each main group, the positioning circuit positions each subgroup within the region corresponding to each main group.

[0006] In some embodiments, the first algorithm may be a deep learning algorithm.

[0007] In some embodiments, the first algorithm may be a triangulation positioning algorithm.

[0008] In some embodiments, the image positioning device further includes a feature extraction circuit. The feature extraction circuit is coupled to the computing circuit and, based on a deep learning algorithm, generates a feature vector corresponding to each image included in each main group as an image feature. The positioning circuit classifies the images included in each main group into multiple subgroups based on the feature vectors of each image included in each main group and the clustering algorithm. The positioning circuit uses an image comparison algorithm to determine the relative positional relationship of each subgroup included in each main group and locates each subgroup based on the relative positional relationship.

[0009] In some embodiments, the image positioning device further comprises a feature extraction circuit. The feature extraction circuit is coupled to the operation circuit. The feature extraction circuit sorts the images included in each master group according to the shooting time, and obtains a first relative movement distance between each image and a previous image included in each master group according to a second algorithm. In addition, the feature extraction circuit obtains a second relative movement distance between each image and a first image included in each master group according to the first relative movement distance corresponding to each image included in each master group, as the feature of the image, wherein the second algorithm can be an image comparison algorithm or an inertial measurement unit algorithm. The positioning circuit classifies the images included in each master group into a plurality of subgroups according to the second relative movement distance of each image included in each master group and the grouping algorithm. The positioning circuit obtains the relative positional relationship of each subgroup included in each master group according to the second relative movement distance of each image included in each master group, and positions each subgroup according to the relative positional relationship.

[0010] In some embodiments, the grouping algorithm can be a k-means algorithm.

[0011] According to an embodiment of the present application, an image positioning method is provided. The image positioning method is applied to an image positioning device. The steps of the image positioning method include: classifying a plurality of images into a plurality of master groups according to a first algorithm by an operation circuit of the image positioning device, wherein each master group corresponds to a different one of the plurality of regions; classifying the images included in each master group into a plurality of subgroups according to the feature of each image included in each master group and a grouping algorithm by the operation circuit; and positioning each subgroup in the region corresponding to each master group according to the relative positional relationship of each subgroup included in each master group by a positioning circuit of the image positioning device.

[0012] As to other additional features and advantages of the present application, those skilled in the art can make minor modifications and embellishments to the image positioning device and method disclosed in the embodiments of the present application without departing from the spirit and scope of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0013] Figure 1 FIG. 1 is a block diagram showing an image positioning device 100 according to an embodiment of the present application.

[0014] Figure 2 FIG. 2 is a schematic diagram showing a plurality of regions included in a positioning environment according to an embodiment of the present application.

[0015] Figure 3FIG. 1 is a schematic diagram of a first relative movement distance and a second relative movement distance of each image included in a main group according to an embodiment of the present invention.

[0016] Figure 4 FIG. 1 is a schematic diagram of a first subgroup and a second subgroup of a main group according to an embodiment of the present invention.

[0017] Figure 5 is a flowchart of an image positioning method according to an embodiment of the present invention.

[0018]

Explanation of symbols

[0019] 100: Image positioning device

[0020] 110: Storage device

[0021] 120: Operational Circuit

[0022] 130: Feature extraction circuit

[0023] 140: Positioning circuit

[0024] 200: Image acquisition device

[0025] I1~I5、P1~P p 、Q1~Q q :image

[0026] A: Positioning environment is divided into

[0027] a1~a5:area

[0028] S510~S540: Steps DETAILED DESCRIPTION

[0029] This section describes the preferred method of implementing the present invention, which is intended to illustrate the spirit of the present invention rather than to limit the scope of protection of the present invention. The scope of protection of the present invention shall be determined by the scope defined in the appended claims.

[0030] Figure 1 FIG. 1 is a block diagram showing an image positioning device 100 according to an embodiment of the present invention. Figure 1 As shown, the image positioning device 100 may include a storage device 110, a computing circuit 120, a feature extraction circuit 130, and a positioning circuit 140. Figure 1 The block diagram shown in FIG is only for the convenience of illustrating the embodiment of the present invention, but the present invention is not limited to Figure 1The image positioning device 100 may also include other components or other connection methods. According to one embodiment of the present invention, the computing circuit 120, the feature extraction circuit 130, and the positioning circuit 140 may be partially or entirely integrated into a single chip. Furthermore, the storage device 110, the computing circuit 120, the feature extraction circuit 130, and the positioning circuit 140 may also be integrated into a single chip.

[0031] According to an embodiment of the present invention, storage device 110 may be a volatile memory (e.g., random access memory (RAM)), a non-volatile memory (e.g., flash memory, read-only memory (ROM)), a hard disk, or a combination thereof. Storage device 110 may be used to store files and data required for positioning, such as information corresponding to multiple areas within a positioning environment, but the present invention is not limited thereto.

[0032] According to an embodiment of the present invention, a positioning environment can be pre-divided into multiple areas, and information corresponding to each area can be pre-stored in the storage device 110. According to an embodiment of the present invention, the positioning environment can be an oral cavity, an indoor space (e.g., an office), or an outdoor space, but the present invention is not limited thereto. Figure 2 FIG. 1 is a schematic diagram of multiple areas included in a positioning environment according to an embodiment of the present invention. Figure 2 As shown, the positioning environment A may be pre-divided into areas a1 to a5, and information corresponding to the areas a1 to a5 may be pre-stored in the storage device 110, but the present invention is not limited thereto.

[0033] According to one embodiment of the present invention, the information corresponding to each area may be information about a wireless access point (AP) configured in each area. In other words, in this embodiment, each area may be configured with a wireless access point, and each wireless access point may have its corresponding information. According to one embodiment of the present invention, the information corresponding to each area pre-stored in storage device 110 may include one or more of the aforementioned embodiments, but the present invention is not limited thereto.

[0034] According to one embodiment of the present invention, an image positioning device 100 can obtain multiple images corresponding to a positioning environment from an image capture device 200 (i.e., images generated by the image capture device 200 capturing different areas of the positioning environment), and perform positioning on the multiple images obtained from the image capture device 200. The following embodiments will provide a more detailed description.

[0035] According to an embodiment of the present application, the image extraction device 200 can be an electronic device with a photographing function, such as a mobile phone, a camera, or a camcorder, but the present application is not limited thereto. According to an embodiment of the present application, the image extraction device 200 can include an inertial measurement unit (IMU) to generate IMU information, such as orientation and angle information, corresponding to each image taken by the image extraction device 200.

[0036] According to an embodiment of the present application, the image extraction device 200 can transmit the extracted images and the information corresponding to each image to the image positioning device 100 through a wireless communication method, such as Bluetooth, WIFI, or mobile communication (cellular network). According to an embodiment of the present application, the information corresponding to each image can include time information corresponding to each image, signal strength information (e.g., Received Signal Strength Indication (RSSI)) received by the image extraction device 200 when each image is taken, and IMU information corresponding to each image, but the present application is not limited thereto.

[0037] According to an embodiment of the present application, when the image positioning device 100 obtains a plurality of images corresponding to a positioning environment from the image extraction device 200, the operation circuit 120 of the image positioning device 100 can classify the plurality of images into a plurality of main groups according to the information corresponding to a plurality of regions stored in the storage device 110 and a first algorithm, wherein each main group corresponds to a region. Figure 2 For example, the operation circuit 120 can classify the plurality of images into images corresponding to regions a1 to a5 according to the stored information corresponding to a plurality of regions and the first algorithm, wherein the images corresponding to the region a1 can be regarded as a first main group, the images corresponding to the region a2 can be regarded as a second main group, the images corresponding to the region a3 can be regarded as a third main group, the images corresponding to the region a4 can be regarded as a fourth main group, and the images corresponding to the region a5 can be regarded as a fifth main group, but the present application is not limited thereto.

[0038] According to an embodiment of the present application, the first algorithm can be a deep learning algorithm. The operation circuit 120 of the image positioning device 100 can determine which region each image in the plurality of images corresponds to according to the machine learning result stored in the storage device 110 and the deep learning algorithm, thereby classifying the plurality of images obtained from the image extraction device 200 into a plurality of main groups.

[0039] According to another embodiment of the present invention, when the information corresponding to the multiple regions includes information corresponding to the wireless access points configured in each region, the first algorithm may be a triangulation positioning algorithm. The computing circuit 120 may classify the multiple images obtained from the image capture device 200 into multiple main groups based on the triangulation positioning algorithm and the signal strength information received by the image capture device 200 when each image was captured (i.e., the signal strength information between the image capture device 200 and different wireless access points).

[0040] According to one embodiment of the present invention, after the computing circuit 120 classifies the multiple images obtained from the image capture device 200 into multiple main groups, the feature extraction circuit 130 of the image positioning device 100 extracts features of the images included in each main group. Next, the computing circuit 120 classifies the images included in each main group into multiple subgroups based on the features of the images included in each main group and a clustering algorithm. This will be described in detail below.

[0041] According to one embodiment of the present invention, the clustering algorithm may be a k-means algorithm, but the present invention is not limited thereto. In the k-means algorithm, the user can predetermine the number of subgroups to be divided into for each main group (i.e., predetermine the number of subgroups within each main group). Furthermore, in the k-means algorithm, each subgroup corresponds to a centroid in the k-means algorithm. In other words, the number of centroids is the same as the number of subgroups.

[0042] According to one embodiment of the present invention, the feature extraction circuit 130 may generate a feature vector corresponding to each image included in each main group according to a deep learning algorithm as a feature of each image. Specifically, the feature extraction circuit 130 may input each image included in each main group into a deep learning algorithm model to obtain a feature vector corresponding to each image. In this embodiment, the deep learning algorithm may be a convolutional neural network (CNN) algorithm (e.g., EfficiebtNet, ResNet, etc.), but the present invention is not limited thereto. In addition, in this embodiment, after the feature extraction circuit 130 obtains the feature vector corresponding to each image included in each main group, the operation circuit 120 may classify the images included in each main group into multiple subgroups according to the feature vector corresponding to each image included in each main group by a clustering algorithm (e.g., k-means algorithm).

[0043] According to another embodiment of the present application, the feature extraction circuit 130 can first sort the images included in each master group according to the shooting time. Then, the feature extraction circuit 130 can obtain a first relative moving distance between each image and the previous image included in each master group according to a second algorithm. Then, in each master group, the feature extraction circuit 130 can obtain a second relative moving distance between each image and the first image according to the first relative moving distance between each image and the previous image, wherein the second relative moving distance between each image and the first image is regarded as the feature of each image.

[0044] For example, Figure 3 Figure 3 is a diagram of the first relative moving distance and the second relative moving distance of each image included in a master group according to an embodiment of the present application. As shown in Figure 3 , the feature extraction circuit 130 can first sort the images I1 to I6 included in a master group according to the shooting time. Then, the feature extraction circuit 130 can obtain the first relative moving distance between the image I1 and the image I2 (e.g., 1 unit moving distance, the moving distance can be pixel, centimeter, meter, etc.), the first relative moving distance between the image I2 and the image I3 (e.g., 2 unit moving distance), the first relative moving distance between the image I3 and the image I4 (e.g., 3 unit moving distance), the first relative moving distance between the image I4 and the image I5 (e.g., 4 unit moving distance), and the first relative moving distance between the image I5 and the image I6 (e.g., 1 unit moving distance) according to the second algorithm. Finally, the feature extraction circuit 130 can obtain the second relative moving distance between the image I1 and the image I1 (i.e., 0), the second relative moving distance between the image I2 and the image I1 (i.e., 1), the second relative moving distance between the image I3 and the image I1 (i.e., 3), the second relative moving distance between the image I4 and the image I1 (i.e., 6), the second relative moving distance between the image I5 and the image I1 (i.e., 10), and the second relative moving distance between the image I6 and the image I1 (i.e., 11) according to the corresponding first relative moving distance of the images I1 to I6. Note that, Figure 3 are only used to illustrate the embodiments of the present application, but the present application is not limited thereto. The first relative moving distance and the second relative moving distance can also be negative numbers.

[0045] According to an embodiment of the present application, the second algorithm can be an image matching algorithm or an inertial measurement unit algorithm, but the present application is not limited thereto. When the second algorithm is the image matching algorithm, the feature extraction circuit 130 can match each image and the previous image included in each master group by the image matching algorithm (e.g., match the image I2 and the image I1 by the image matching algorithm, match the image I3 and the image I2 by the image matching algorithm, match the image I4 and the image I3 by the image matching algorithm, match the image I5 and the image I4 by the image matching algorithm, and match the image I6 and the image I5 by the image matching algorithm). Figure 3 ​When the second algorithm is an inertial measurement unit algorithm, the feature extraction circuit 130 can use the inertial measurement unit algorithm to calculate the acceleration value corresponding to each image included in each main group and the previous image (for example, to obtain Figure 3 The acceleration values ​​corresponding to the image I2 and the image I1 are obtained to obtain the first relative movement distance of each image.

[0046] Furthermore, in this embodiment, after the feature extraction circuit 130 obtains the second relative movement distance corresponding to each image, the operation circuit 120 may classify the images included in each main group into a plurality of subgroups using a clustering algorithm (e.g., a k-means algorithm) based on the second relative movement distance corresponding to each image included in each main group.

[0047] According to an embodiment of the present invention, after the computing circuit 120 classifies the images contained in each main group into multiple subgroups, the positioning circuit 140 calculates the relative positional relationships of each subgroup within each main group and locates each subgroup within the region corresponding to each main group based on the relative positional relationships of each subgroup within each main group. A more detailed description will be provided below.

[0048] According to one embodiment of the present invention, when the computing circuit 120 classifies the images contained in each main group into a plurality of subgroups using a clustering algorithm (e.g., a k-means algorithm) based on the feature vectors corresponding to each image contained in each main group, the positioning circuit 140 can calculate the relative position relationship of each subgroup contained in each main group using an image comparison algorithm (or an inertial measurement unit algorithm). In this embodiment, the relative position relationship can be a horizontal position relationship and / or a vertical position relationship. Figure 4 In this embodiment, after the positioning circuit 140 knows the relative position relationship of each sub-group included in a main group, the positioning circuit 140 can locate each sub-group in the area corresponding to the main group.

[0049] Figure 4 FIG is a schematic diagram of a first subgroup and a second subgroup of a main group according to an embodiment of the present invention. Figure 4 As shown, the first subgroup includes p images and the second subgroup includes q images. The positioning circuit 140 can calculate the first image P1 of the first subgroup and all images Q1-Q of the second subgroup by using an image comparison algorithm. q The moving distance h 1，1 , h 1，2 …h 1，q , where h 1，1This means the moving distance between the first image P1 of the first subgroup and the first image Q1 of the second subgroup. 1，1 , h 1，2 …h 1，q Take the average to obtain the first image P1 of the first subgroup and all images Q1~Q q An average moving distance m1 (ie m1=(h 1，1 +h 1，2 +…+h1, q ) / q). Similarly, the positioning circuit 140 can calculate each image P1-P of the first subgroup. p and all images Q1~Q of the second subgroup q The average moving distance m1, m2…m p Then, the positioning circuit 140 can calculate the average moving distances m1, m2, ...m p Take the average value to obtain an average value r (ie r=(m1+m2+…+m p ) / p). The positioning circuit 140 can determine the relative position relationship between the first subgroup and the second subgroup based on the average value r. Specifically, when the average value r is greater than 0, it means that the first subgroup is to the right (or above) the second subgroup, and when the average value r is less than 0, it means that the first subgroup is to the left (or below) the second subgroup. Note that, Figure 4 This is only used to illustrate the embodiments of the present invention, but the present invention is not limited thereto.

[0050] According to another embodiment of the present invention, when the computing circuit 120 classifies the images contained in each main group into a plurality of subgroups based on the second relative movement distance corresponding to each image contained in each main group by using a clustering algorithm (e.g., a k-means algorithm), the positioning circuit 140 can obtain the relative position relationship of each subgroup contained in each main group based on the second relative movement distance of each image contained in each main group, and locate each subgroup based on the relative position relationship. In this embodiment, the relative position relationship can be a horizontal position relationship and / or a vertical position relationship. Figure 3 In this embodiment, after the positioning circuit 140 knows the relative position relationship of each sub-group included in a main group, the positioning circuit 140 can locate each sub-group in the area corresponding to the main group.

[0051] like Figure 3As shown, assuming that image I1, image I2 and image I3 are the first subgroup, image I4 is the second subgroup, and image I5 and image I6 are the first subgroup, the positioning circuit 140 can determine the relative position relationship between each subgroup and image I1 based on the average second relative movement distance of each subgroup, so as to know the relative position relationship of the first subgroup, the second subgroup and the third subgroup contained in the main group. Specifically, when the average second relative movement distance is positive, the positioning circuit 140 can determine that the subgroup is to the right (or above) of image I1, and when the average second relative movement distance is negative, the positioning circuit 140 can determine that the subgroup is to the left (or below) of image I1. In addition, when the average second relative movement distance is close to 0, the positioning circuit 140 can determine that the subgroup is close to image I1, and when the average second relative movement distance 0 is farther, the positioning circuit 140 can determine that the subgroup is farther away from image I1. Note that, Figure 3 This is only used to illustrate the embodiments of the present invention, but the present invention is not limited thereto.

[0052] Figure 5 FIG. 1 is a flow chart of an image positioning method according to an embodiment of the present invention. The image positioning method can be applied to the image positioning device 100. Figure 5 As shown, in step S510 , a computing circuit of the image positioning device 100 obtains a plurality of images from an image capturing device.

[0053] In step S520 , the computing circuit of the image positioning device 100 classifies the plurality of images into a plurality of main groups according to a first algorithm and information corresponding to the plurality of regions stored in a storage device of the image positioning device 100 , wherein each main group corresponds to a different one of the plurality of regions.

[0054] In step S530 , the computing circuit of the image positioning apparatus 100 classifies the images included in each main group into a plurality of subgroups according to the features of each image included in each main group and a clustering algorithm.

[0055] In step S540 , a positioning circuit of the image positioning device 100 positions each subgroup in the area corresponding to each main group according to the relative position relationship of each subgroup included in each main group.

[0056] According to some embodiments of the present invention, in the image positioning method, the first algorithm is a deep learning algorithm. The computing circuit of the image positioning device 100 can classify the plurality of images into a plurality of main groups according to the deep learning algorithm.

[0057] According to some embodiments of the present invention, in an image positioning method, the information corresponding to the multiple regions includes information about a wireless access point configured for each of the multiple regions, and the first algorithm is a triangulation positioning algorithm. In the image positioning method, the computing circuit of the image positioning device 100 can classify the multiple images into multiple main groups based on the signal strength corresponding to the wireless access point corresponding to each of the multiple regions and the triangulation positioning algorithm.

[0058] According to some embodiments of the present invention, the image positioning method further includes a feature extraction circuit of the image positioning device 100 generating a feature vector corresponding to each image included in each main group according to a deep learning algorithm as a feature of the image. In these embodiments, the positioning circuit of the image positioning device 100 classifies the images included in each main group into multiple subgroups based on the feature vector of each image included in each main group and the clustering algorithm. In these embodiments, the positioning circuit of the image positioning device 100 determines the relative positional relationship of each subgroup included in each main group according to an image comparison algorithm, and locates each subgroup based on the relative positional relationship of each subgroup.

[0059] According to some embodiments of the present invention, the image positioning method further includes steps in which a feature extraction circuit of the image positioning device 100 sorts the images included in each main group by capture time. Next, the feature extraction circuit of the image positioning device 100 obtains a first relative movement distance between each image included in each main group and the previous image based on a second algorithm. Next, based on the first relative movement distance corresponding to each image included in each main group, the feature extraction circuit of the image positioning device 100 obtains a second relative movement distance between each image included in each main group and the first image included in each main group as an image feature.

[0060] In these embodiments, the second algorithm may be an image comparison algorithm or an inertial measurement unit algorithm. In these embodiments, the positioning circuit of the image positioning device 100 classifies the images included in each main group into a plurality of subgroups based on the second relative movement distance of each image included in each main group and the grouping algorithm.

[0061] In these embodiments, the positioning circuit of the image positioning device 100 obtains the relative position relationship of each subgroup included in each main group according to the second relative movement distance of each image included in each main group, and positions each subgroup according to the relative position relationship of each subgroup.

[0062] According to some embodiments of the present invention, the clustering algorithm may be a k-means algorithm.

[0063] The positioning method proposed in this invention first categorizes multiple images into distinct main groups across a large area, then further categorizes the subgroups within each main group, and locates each subgroup based on their relative positions. Therefore, the positioning method proposed in this invention allows for more precise spatial location determination of multiple images.

[0064] Serial numbers in this specification and claims, such as "first", "second", etc., are only for convenience of description and have no sequential relationship with each other.

[0065] The methods and algorithm steps disclosed in the present invention can be directly implemented in hardware and software modules, or a combination of both, by executing a processor. A software module (including execution instructions and associated data) and other data can be stored in a data storage device, such as random access memory (RAM), flash memory, read-only memory (ROM), erasable programmable read-only memory (EPROM), electronically erasable programmable read-only memory (EEPROM), a temporary register, a hard drive, a removable hard drive, a compact disc read-only memory (CD-ROM), a DVD, or any other computer-readable storage medium format known in the art. A storage medium can be coupled to a machine device, such as a computer / processor (referred to as a processor in this specification for ease of explanation), so that the processor can read information (such as program code) and write information to the storage medium. A storage medium can be integrated with a processor. An application-specific integrated circuit (ASIC) includes a processor and a storage medium. A user device includes an ASIC. In other words, the processor and storage medium are included in the user device without being directly connected to the user device. Furthermore, in some embodiments, any suitable computer program product may include a readable storage medium, wherein the readable storage medium includes program code related to one or more disclosed embodiments. In some embodiments, the computer program product may include packaging materials.

[0066] The above paragraphs describe various aspects. Obviously, the teachings herein can be implemented in a variety of ways, and any specific architecture or functionality disclosed in the examples is merely representative. Based on the teachings herein, those skilled in the art will understand that each aspect disclosed herein can be implemented independently or in combination with two or more aspects.

[0067] Although the present disclosure has been disclosed above with reference to the embodiments, they are not intended to limit the present disclosure. Any person skilled in the art may make slight changes and modifications without departing from the spirit and scope of the present disclosure. Therefore, the scope of protection of the invention shall be determined by the scope defined in the appended claims.

Claims

1. An image positioning device, comprising: an arithmetic circuit, the arithmetic circuit acquiring a plurality of images from the image capture device and classifying the plurality of images into a plurality of main groups based on a first algorithm, wherein each main group corresponds to a different area of ​​the positioning environment, and the arithmetic circuit classifying the images contained in each main group into a plurality of subgroups based on features of each image contained in the main group and the clustering algorithm; as well as a positioning circuit coupled to the arithmetic circuit, wherein the positioning circuit positions each subgroup in the region corresponding to each main group based on a relative positional relationship between each subgroup included in each main group; A feature extraction circuit is coupled to the arithmetic circuit, wherein the feature extraction circuit sorts the images included in each main group according to shooting time and obtains a first relative movement distance between each image included in each main group and a previous image according to a second algorithm. The feature extraction circuit also obtains a second relative movement distance between each image included in each main group and a first image included in each main group based on the first relative movement distance corresponding to each image included in each main group, as the feature of the image. 2 . The image positioning device as claimed in claim 1 , wherein the first algorithm is a deep learning algorithm or a triangulation positioning algorithm.

3. The image positioning device according to claim 1, further comprising: The feature extraction circuit is coupled to the operation circuit and generates a feature vector corresponding to each image included in each main group according to a deep learning algorithm to serve as the feature of the image.

4. The image positioning device according to claim 3, wherein the positioning circuit classifies the images included in each main group into the plurality of subgroups based on the feature vector of each image included in each main group and the clustering algorithm.

5. The image positioning device of claim 4, wherein the positioning circuit obtains the relative position relationship of each of the subgroups included in each of the main groups according to an image comparison algorithm, and positions each of the subgroups according to the relative position relationship. 6 . The image positioning device as claimed in claim 1 , wherein the second algorithm is an image comparison algorithm or an inertial measurement unit algorithm.

7. The image positioning device of claim 1, wherein the positioning circuit classifies the images included in each main group into the plurality of subgroups based on the second relative movement distance of each image included in each main group and the clustering algorithm.

8. The image positioning device of claim 7 , wherein the positioning circuit obtains the relative positional relationship of each subgroup included in each main group based on the second relative movement distance of each image included in each main group, and positions each subgroup based on the relative positional relationship.

9. The image positioning device as claimed in claim 1, wherein the clustering algorithm is a k-means algorithm.

10. An image positioning method, applicable to an image positioning device, comprising: Acquire a plurality of images from an image capture device by means of the computing circuit of the image positioning device; The computing circuit classifies the plurality of images into a plurality of main groups according to a first algorithm, wherein each main group corresponds to a different area of ​​the positioning environment; Classifying the images included in each main group into a plurality of subgroups by the computing circuit according to the features of each image included in each main group and a clustering algorithm; and Using a positioning circuit of the image positioning device, each subgroup is positioned in the region corresponding to each main group according to the relative positional relationship of each subgroup included in each main group; The images contained in each main group are sorted according to shooting time by the feature extraction circuit of the image positioning device; Obtaining a first relative movement distance between each image included in each main group and a previous image by the feature extraction circuit according to a second algorithm; and The feature extraction circuit obtains the second relative movement distance between each image included in each main group and the first image included in each main group according to the first relative movement distance corresponding to each image included in each main group, as the feature of the image.

11. The image positioning method as claimed in claim 10, wherein the first algorithm is a deep learning algorithm or a triangulation positioning algorithm.

12. The image positioning method according to claim 10, further comprising: The feature extraction circuit of the image positioning device generates a feature vector corresponding to each image included in each main group according to a deep learning algorithm to serve as the feature of the image.

13. The image positioning method according to claim 12, further comprising: The positioning circuit classifies the images included in each main group into the plurality of subgroups according to the feature vector of each image included in each main group and the clustering algorithm.

14. The image positioning method according to claim 13, further comprising: The positioning circuit obtains the relative position relationship of each of the subgroups included in each of the main groups according to an image comparison algorithm, and positions each of the subgroups according to the relative position relationship. 15 . The image positioning method as claimed in claim 10 , wherein the second algorithm is an image comparison algorithm or an inertial measurement unit algorithm.

16. The image positioning method according to claim 10, further comprising: The positioning circuit classifies the images included in each main group into the plurality of subgroups according to the second relative movement distance of each image included in each main group and the clustering algorithm.

17. The image positioning method according to claim 16, further comprising: The positioning circuit obtains the relative positional relationship of each subgroup included in each main group according to the second relative movement distance of each image included in each main group, and positions each subgroup according to the relative positional relationship.

18. The image localization method according to claim 10, wherein the clustering algorithm is a k-means algorithm.

Images